test: cover FME rollout prompt, governance CRUD scopes, infra deep links

[cursor]

Description

Adds regression tests for recently merged production paths that lacked CRUD/prompt-level coverage:

• tests/prompts/feature-flag-rollout.test.ts — dual-mode scope validation (workspaceId vs orgId+projectId) and Harness-native NYI caveat for the FME rollout prompt (feat: [FME-17249]: FME dual-mode Harness-native scope migration #806)
• tests/registry/governance.test.ts — get/create/update/delete resource_scope param injection for policy and policy_set at account/org/project (feat: allow account and org scope for OPA policies and policy sets #819)
• tests/registry/infrastructure.test.ts — nested environmentRef aliasing and environment_id param fallback for openInHarness deep links (feat: [AIPLAT-1334]: Point infrastructure openInHarness at environment Infrastructure section #809)

Type of Change

• Bug fix
• New feature
• Refactor
• Documentation
• Other — test coverage

Checklist

• pnpm test passes (78 targeted tests)
• pnpm typecheck passes
• pnpm build passes
• pnpm standards:check passes
• pnpm docs:check passes

Risky behavior now covered

1. FME rollout prompt scope gate — agents calling the prompt without valid scope get a clear error instead of generating broken rollout instructions; Harness-native mode surfaces the NYI caveat for steps that still require legacy workspace_id.
2. Governance multi-scope CRUD — account/org/project scope params are correctly injected (or omitted) on get/create/update/delete, preventing silent wrong-scope API calls.
3. Infrastructure deep links — list/get responses with nested environmentRef or caller-supplied environment_id still produce correct Harness UI links.

Why these tests materially reduce regression risk

These paths are high-blast-radius: the FME dual-mode migration and governance scope expansion changed shared dispatch behavior. A regression would misroute API calls or emit broken deep links without obvious runtime errors. The tests lock the request-shape contracts and prompt validation that production code now depends on.

  

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[cursor]

PR Risk Assessment — Very Low

Evidence (diff-only)

File	Change
tests/prompts/feature-flag-rollout.test.ts	New (+123) — prompt scope validation & dual-mode tests
tests/registry/governance.test.ts	Modified (+90) — CRUD scope param injection tests
tests/registry/infrastructure.test.ts	Modified (+46) — deep-link fallback/alias tests

Total: 3 files, +259 / −0. No src/ or production code changes.

Risk factors

• Codepaths: Test suite only; exercises existing prompt, registry dispatch, and deep-link extractors via mocks.
• Blast radius: None — no runtime behavior changes.
• Complexity: Straightforward Vitest cases with in-memory MCP transport and mocked HTTP client.
• Infra / security / user-facing: No impact.

Decision

Item	Result
Risk level	Very Low
Review required	No
Reviewers assigned	None (not warranted at this level)
Approval	Already approved by @puthrayaharness on 5b571aa — no additional approval from this automation
CODEOWNERS	None configured in this repo

CI checks (build-and-test, smoke tests) are green. This PR adds regression coverage for recently merged production paths and is safe to merge from a risk perspective.

  

_{Sent by Cursor Automation: Assign PR reviewers}

[cursor]

No inline review comments were found on this PR to address.

Checked via GitHub API (pull review comments, review threads, and issue comments). The only review is an approval from @puthrayaharness with no inline feedback. No code changes were made.

  

_{Sent by Cursor Automation: Autofix PR review comments}

[thisrohangupta]

Closing as superseded. This PR duplicates the older FME rollout/governance/infrastructure coverage and encodes the pre-native-support caveat. The current coverage and cleanup are being consolidated in #842, with the small source cleanup from #845 folded there.